remove local_mul_to_neg as it doing a subset of local_mul_specialize

theano/tensor/opt.py

@@ -3385,33 +3385,6 @@ def local_sum_alloc(node):
                     pass
 
 
-@gof.local_optimizer([T.mul])
-def local_mul_to_neg(node):
-    """
-    mul(-1, x) -> neg(x)
-
-    This is not done if we would add more nodes in the graph, like with:
-
-    mul(-1, x, y) -/-> neg(mul(x, y))
-    """
-    if (node.op == T.mul and
-        local_mul_canonizer.get_constant(node.inputs[0]) == -1.0):
-        num = node.inputs[1:]
-        if len(num) == 1:
-            other_prod = num[0]
-        else:
-            # This would add extra node in the graph
-            return False
-            other_prod = local_mul_canonizer.main(*num)
-        if other_prod.type == node.outputs[0].type:
-            return [-other_prod]
-        # else the multiplication is also acting as a cast, so we
-        # might as well leave it alone.  I don't think it's better to
-        # turn this into a negation in the wrong type, followed by an
-        # explicit cast.
-register_specialize(local_mul_to_neg)
-
-
 @register_specialize
 @gof.local_optimizer([T.neg])
 def local_neg_neg(node):

theano/tensor/tests/test_opt.py

@@ -4000,27 +4000,6 @@ def test_local_join_1():
     assert f.maker.fgraph.outputs[0].dtype == config.floatX
 
 
-def test_local_mul_to_neg():
-    """
-    Test that a multiplication by -1 or -1.0 yields the appropriate data type
-    """
-    a = T.imatrix()
-    f1 = theano.function([a], -1 * a)
-    f2 = theano.function([a], -1.0 * a)
-    aval = numpy.random.randint(0, 10, (2, 2)).astype('int32')
-    if config.cast_policy == 'custom':
-        assert f1(aval).dtype == a.dtype
-        assert f2(aval).dtype == 'float64'
-    elif config.cast_policy == 'numpy':
-        assert f1(aval).dtype == str(numpy.array(0).dtype)
-        assert f2(aval).dtype == 'float64'
-    elif config.cast_policy == 'numpy+floatX':
-        assert f1(aval).dtype == str(numpy.array(0).dtype)
-        assert f2(aval).dtype == config.floatX
-    else:
-        raise NotImplementedError(config.cast_policy)
-
-
 def test_local_add_specialize():
     # test of non-zero dimension
     a = tensor.vector()